Anticipating Critical Events to Customise and Improve the Performance of Federation Runtime Infrastructure

this paper proposes a methodology to improve the time-performance of distributed simulation based on the HLA architecture using commercial simulators. In the context of industrial production, for instance, the timing of information transfer among federate simulators becomes an important issue whenever production control and scheduling are very closely linked. In such a context the re-scheduling of an existing production plan, as determined for instance by a rush order, is usually decided upon at the control level in the hierarchy of federates; however it cannot be communicated to the different federate plants until a status/parameter update is instantiated and established. The study presented in this paper specifically focuses on the impact of machine failures at the federate plant level. The random nature of the failure processes associated to the machines of the federate production plants makes it difficult to establish when such occurrences may take place: therefore the events driving the exchange of state variables among federates cannot be set with certainty ahead of time. This research proposes a one-off methodology to statistically determine when critical events, such as machine failures, are likely to occur. Based on such information it is possible to intersperse additional events among those associated to the usual production/process milestones and thus significantly reduce delays in the transfer of critical status parameters among federates and waste of simulated time.